def task_gen_top_tag_via_number(query, K, root):
from photo_dao import PhotoDao
from database import DBHelper
db_helper = DBHelper()
db_helper.init(root)
photo_dao = PhotoDao(db_helper)
photo_ids = photo_dao.getClassPhotoIds(query, ''.join([query]))
photos = photo_dao.getPhotos(query, photo_ids)
hist = {}
for photo in photos:
tags = photo.tags
for tag in tags:
if(tag in hist):
hist[tag] = hist[tag] + 1
else:
hist[tag] = 0
top_word_freq = sorted(hist.items(), key=lambda t: -t[1])
top_word_freq = top_word_freq[0:min(len(top_word_freq), K)]
top_word = []
for line in top_word_freq:
top_word.append(line[0].strip())
output_path = ''.join([root, '/data/tags/%s.txt'%query])
from file_io import FileIO
file_io = FileIO()
file_io.write_strings_to_file(top_word, output_path)
return top_word
def task_gen_top_tag_via_user(photos, query, K, root, do_skip):
tags_path = ''.join([root, '/data/tags/%s.txt' % query])
import os
if os.path.exists(tags_path) and do_skip:
print ('top tags already generated.\n')
fin = open(tags_path, 'r')
top_word = []
for line in fin:
top_word.append(line.strip())
return top_word
print('begin gen top tag via user\n')
tag_user_ids = {}
for photo in photos:
tags = photo.tags
for tag in tags:
if (not(tag in tag_user_ids)):
tag_user_ids[tag] = ''
tag_user_ids[tag] = ''.join([tag_user_ids[tag], ',', photo.ownerId])
tags_score = {}
for tag in tag_user_ids:
user_ids = tag_user_ids[tag]
user_ids = user_ids.split(',')
tags_score[tag] = len(user_ids)
top_word_freq = sorted(tags_score.items(), key=lambda t: -t[1])
K = min(len(top_word_freq), K)
top_word_freq = top_word_freq[0:K]
top_word = []
for line in top_word_freq:
top_word.append(line[0].strip())
print('done gen top tag via user, saving output ...\n')
# print out top tags.
from file_io import FileIO
file_io = FileIO()
file_io.write_strings_to_file(top_word, tags_path)
# print out score for tags.
top_word_str = []
for line in top_word_freq:
top_word_str.append(line[0].strip() + ' ' + str(line[1]))
output_path = ''.join([root, '/data/tags/%s_score.txt'%query])
from file_io import FileIO
file_io = FileIO()
file_io.write_strings_to_file(top_word_str, output_path)
save_tag_to_mat(top_word, K, query, root)
return top_word
def task_gen_photo_imagepath(root, query):
print('Get photo ids.');
db_helper = DBHelper();
db_helper.init(root);
photo_dao = PhotoDao(db_helper);
tic();
photo_ids = photo_dao.getClassPhotoIds(query, ''.join([query]));
toc();
print('Get photo path.');
imagepaths = get_photo_imagepath(root, query, photo_ids)
output_path = ''.join([db_helper.datasetDir, '/', query, '_imagepath.txt']);
file_io = FileIO();
file_io.write_strings_to_file(imagepaths, output_path);
